The Global Trade Analysis Project

1. The Global Trade Analysis Project Presented by Badri Narayanan G. Center for Global Trade Analysis Purdue University

2. Outline • The Project • The Network • The Data Base • The Model • Other GTAP Offerings • An Illustration of GTAP: The PE-GE model

3. The Project • International network of economic/policy researchers and policy-makers • Quantitative analysis of international policy issues within an economy-wide framework • Co-ordination: Center for Global Trade Analysis, Purdue University • Funding: • Consortium subscriptions • Data Base Sales • Project-based 3

4. The Network • 31 Consortium Board Members (US Govt agencies, EC, OECD, WB, UN, WTO, etc.) • Over 9000 network members from 159 countries • Over 2000 contributing members from 99 countries 4 4

5. The Network

6. GTAP Data Base • Multi-country multi-sector economic data • Input-Output (I-O) data • Trade data: imports, exports, margins. • Macro-economic data: GDP, population, consumption, investment, government expenditure, savings… • Energy data: volumes, taxes, prices • Protection data: taxes, domestic support, subsidies, tariffs, preferences, quotas… • Satellite data: Land-use, Emissions, Bio-fuels...

7. History

8. How can you use GTAP Data? • International Computable General Equilibrium Modeling: Some GTAP Applications • Global Trade Analysis: FTAs, PTAs, WTO, etc. • Poverty • Energy, Environment and climate change • Land Use and agricultural issues • Migration and labor issues • Growth and development • Factor movements and technology

9. How is GTAP Data Base constructed? • Collect the I-O tables from contributors • “Process” them • Reconcile them with international datasets • Assemble the data into a single consistent and balanced database

10. I-O Data Final demand Production Domestic inputs Imported inputs Primary factors

11. I-O USA I-O JPN I-O REG N Trade I-O REG 4 I-O IND Data File Content

12. Construction Process I-O Tables (Processed) FIT International Data Sets Fitted I-O Tables Assemble GTAP Data Base

13. Clean, Disaggregate, Synthesize • Clean • Remove small remaining problems with balance and sign. • Disaggregate • Of the 113 regions in GTAP 7: only 35 I-O tables have all 57 sectors; no disaggregation needed • 40 tables need agricultural disaggregation; use agricultural I-O data set. • 17 tables need non-agricultural disaggregation; use representative table. • Synthesize • Create 19 composite regions.

14. FIT: Updating and Reconciliation • Eliminate changes in stocks • Reconcile with international data sets • Entropy theoretic approach

15. International Data Sets in GTAP agricultural data set macro data set trade data set Protection data set energy data set

16. Data Assembly Income / Factor Taxes FIT'ed I-O Tables Parameters Primary Factor Splits Assemble GTAP Data Base

17. Data Base Modification Tools • Beginners: ViewHAR, GTAPAgg, FlexAgg • GAMS users: HAR2GDX, GDX2HAR • Advanced users: • SplitCom , MSplitcom– preserve the overall balance in the data while disaggregating the sectors using specified weights. • Altertax – tax rate/tariff changes while preserving the balance, using a specific GE closure with appropriate elasticities with least possible changes to other data

18. Main Data Construction Program • one job with run time: 10 hours • 215 data handling programs, 17k data files • 22 top level modules: trade, energy, etc. • with sub-modules: 38 modules total • Make: build management tool to keep outputs up to date with inputs and programs • error handling inbuilt

19. Formats and Software used • initial data: HAR, GEMPACK text • internal data: HAR • data handling: TABLO • text processing: Unix utilities • miscellaneous: Bash, GAMS

20. GTAP Model • Hertel (1997) • Elasticities: calibrated/estimated econometrically • Demand = Supply in all markets (price = marginal cost) • Taxes: wedge between prod & cons prices • Int’l trade: Armington CES substitution across sources • Firm Production Inputs: intermediate & factors: Leontief • Intermediate: imported/domestic: CES • Factors: Labor, Capital, Land: CES • Regional Household: Y=C+I+G+X-M: constant shares • Private HHLD (C): CDE demand system: Hanoch (1975) • Global savings (Fixed share of income, ROR): investment • Welfare Decomposition: EV

21. Other GTAP Offerings • GTAP Model and data extensions: Land use, agriculture, bio-fuels, dynamic, migration, climate change, poverty, imperfect competition • Short courses • Conferences • Online resources • Mailing list: GTAP-l • Other models/utilities: GTAPinGAMS, FTAP, CRUSOE, etc.

22. An Illustration: the PE-GE model • Tariff Variations at Disaggregate Level Tariff aggregation problems • “False competition” • So, disaggregated Partial Equilibrium (PE) models are used as inputs to negotiations.Is PE enough? What about welfare gains and economy-wide effects? • We develop a GTAP PE-GE model with provisions to do tariff analysis at disaggregate level. • Indian auto industry is a good example: heterogenous, divergent tariffs, contentious tariff cut proposals. 22

23. An Illustration: the PE-GE model • Model Summary • CET and CES nests used to aggregate supply and demand, respectively; corresponding price linkages. • Armington nest and CES nest between domestic and import demand: based on GTAP model. • Market Clearing to determine Market Prices • Transport Margins: Based on GTAP model • Welfare Decomposition extended to the disaggregated level: AE and TOT Effects • Slack Variables used to link GE and PE parts 23

24. An Illustration: the PE-GE model • Database • GTAP Data Base version 6.2 • TASTE software for MacMAP HS6-level trade and tariff data, mapped to GTAP aggregation • Tariff adjustments in GTAP to accommodate MacMAP: Altertax simulation • Regions: • India (IND) • East and South East Asia (SEA) • Rest of the World (ROW) 24

25. An Illustration: the PE-GE model Database (Contd…) Sectors TRAD_COMM 1. Food (food) 2. Sectors that supply Raw Materials to Auto (autorms) 3. Energy (energy) 4. Auto (AUTO): a. Motor Cycles (MCYC) b. Motor Cycle Parts (MCYP) c. Automobiles other than motorcycles (ATML) d. Engines and other Parts of Automobiles (ATMP) e. Other Transport Equipment (OTHR) Disaggregated Sector DAGG_COMM Sub-Sectors SSECT_COMM 25

26. The Database (Contd.) 26

27. The Database (Contd.) • Some Inferences from the database: • Divergent tariffs: Highest for Automobiles, MCs • All tariffs for imports from SEA are higher than for those from ROW, most divergent for OtherTrans, Automobile, MC. • OtherTrans (ROW) and Engines&Parts (SEA) dominate India’s Auto imports • SEA’s share in India’s total imports is lower for AUTO than in MCs, MC parts and Engines&Parts

28. Closure 28

29. Results Imports from ROW 29

30. Results (Contd.) Imports from SEA 30

31. Results (Contd.) Results from Systematic Sensitivity Analysis

32. Results (Contd.) Welfare: Aggregate Results 32

33. Results (Contd.) Welfare: Import-tax-related Results in Indian Auto Sector

34. Conclusions • PE Model captures sub-sector info but ignores economy-wide impacts  Huge price adjustments, little quantity changes! • GE model ignores sub-sector-level details. • PE-GE results: closer to GE, but quite different. • Auto imports by India rise sharper in PE-GE • Heavy influx of automobiles and Mcycles! 34

35. Conclusions (Contd…) • “False Competition” in some sub-sectors Substitution from ROW to SEA: lesser extent in PE-GE, as SEA has a lower share in India’s AUTO imports, but not at SSECT level. • Welfare differences are notable: • India’s net welfare loss is much lower in PE-GE • India loses more in TOT and gains more! • Results not sensitive to assumed elasticities.